OKR: writing objectives and key results that survive the quarter

OKR stands for objectives and key results, a goal-setting method in which a short qualitative objective is paired with a few measurable key results that define what achieving it looks like. The objective is the direction; the key results are how you would know you got there.

The method is simple enough to explain in a paragraph and is nonetheless abandoned by most teams that try it. The reason is almost always the same: the objectives were vague, the key results were a list of things somebody planned to do anyway, and nothing about the quarter changed as a result. A well-written set does change something — it makes explicit what will be given up.

What a good pair looks like

• The objective is short, qualitative and memorable. Somebody should be able to repeat it without reading it. Make onboarding something new customers finish without help beats Improve onboarding KPIs.
• Two to four key results per objective. More than four means the objective is not one objective.
• Key results are outcomes, not activities. Ship the new signup flow is a task; 70% of new accounts complete setup without contacting support is a key result.
• Each key result has a number and a direction, from a known starting point. From 45% to 70% is a target; improve is a wish.
• The set is honest about the cost. If achieving the objective means something else gets less attention, say what.

The failure modes, in order of frequency

1. Task lists dressed as key results. The test: could you complete all of them and still not have achieved the objective? If yes, they are activities.
2. Too many objectives. Three per team per quarter is a lot. Five is a way of avoiding a decision about priorities.
3. Metrics nobody controls. A key result tied to something the team cannot influence teaches everyone that the exercise is theatre.
4. Sandbagging, which happens whenever OKRs are tied to compensation. Targets set to be hit are not targets, and this is the single most common way the method is destroyed — covered from the other side in performance review.
5. Set and forgotten. Without a short weekly check the whole thing is a document written in January and discovered in April.

Keep OKRs away from pay. The moment achieving them determines a bonus, people negotiate targets they know they can hit, and the method stops describing ambition and starts describing certainty. That is a rational response to the incentive, not a character flaw, and the only fix is to separate the two systems.

A workable quarterly rhythm

• Draft at the end of the previous quarter, not two weeks into the new one.
• Write them with the people doing the work rather than for them. Handed-down OKRs get compliance; co-written ones get judgement.
• Check weekly for five minutes: confidence up or down, and why. Not a status report.
• Mid-quarter, allow one honest revision. Circumstances change and defending an irrelevant target for six more weeks helps nobody.
• Close the quarter by scoring and, more importantly, by writing down what you learned. The score is less useful than the explanation.

Whether you need them at all

OKRs suit organisations big enough that teams can drift apart without noticing — roughly the point where not everyone is in the same conversation. Below that, a team of five running quarterly OKRs is usually adding ceremony to something they already coordinate by talking. The honest alternatives are a short written plan with owners and dates, or the operational numbers you already watch, discussed in kpi dashboard and balanced scorecard.

Where they live

Ettex Board carries the objectives and the work underneath them with owners and dates, so the weekly check reads from the same place the work happens; the written set and its reasoning belong in Ettex Docs, and any numeric key results usually come from Ettex Sheets.

Being direct: this is not OKR software. There is no scoring engine, no alignment tree, no cascading from company to team to individual. For a business under a few hundred people that machinery tends to cost more attention than it returns, and a written page plus a board covers what the method actually requires.